Geographical structure, narrow species ranges, and Cenozoic diversification in a pantropical clade of epiphyllous leafy liverworts

Abstract The evolutionary history and classification of epiphyllous cryptogams are still poorly known. Leptolejeunea is a largely epiphyllous pantropical liverwort genus with about 25 species characterized by deeply bilobed underleaves, elliptic to narrowly obovate leaf lobes, the presence of ocelli, and vegetative reproduction by cladia. Sequences of three chloroplast regions (rbcL, trnL‐F, psbA) and the nuclear ribosomal ITS region were obtained for 66 accessions of Leptolejeunea and six outgroup species to explore the phylogeny, divergence times, and ancestral areas of this genus. The phylogeny was estimated using maximum‐likelihood and Bayesian inference approaches, and divergence times were estimated with a Bayesian relaxed clock method. Leptolejeunea likely originated in Asia or the Neotropics within a time interval from the Early Eocene to the Late Cretaceous (67.9 Ma, 95% highest posterior density [HPD]: 47.9–93.7). Diversification of the crown group initiated in the Eocene or early Oligocene (38.4 Ma, 95% HPD: 27.2–52.6). Most species clades were established in the Miocene. Leptolejeunea epiphylla and L. schiffneri originated in Asia and colonized African islands during the Plio‐Pleistocene. Accessions of supposedly pantropical species are placed in different main clades. Several monophyletic morphospecies exhibit considerable sequence variation related to a geographical pattern. The clear geographic structure of the Leptolejeunea crown group points to evolutionary processes including rare long‐distance dispersal and subsequent speciation. Leptolejeunea may have benefitted from the large‐scale distribution of humid tropical angiosperm forests in the Eocene.


| INTRODUCTION
Range estimation is a challenging theme in morphologically little differentiated groups of organisms and suitable to improve understanding of species diversity and evolution. Many bryophyte genera belong to these critical groups and are in need of thorough reinvestigation including integrative molecular-morphological approaches; however, to date, only a limited number of studies is available (Dong et al., 2012;Forrest, Salazar-Allen, Gudiño, Korpelainen, & Long, 2011;Hedenäs et al., 2014;Heinrichs et al., 2015;Renner et al., 2013;Vanderpoorten, Patiño, Dirkse, Blockeel, & Hedenäs, 2015;Vigalondo et al., 2016). These studies identified numerous morphologically not or weakly differentiated bryophyte species of which many have rather narrow ranges.
Leptolejeunea (Spruce) Steph. is a pantropical genus of nearly exclusively epiphyllous leafy liverworts that grow in lowland and lower montane rainforests, occasionally also in high montane rainforests up to ca. 3,000 m (Bischler, 1969). The genus includes both local endemics (Shu, Zhu, & Pócs, 2016) and intercontinentally distributed species such as L. elliptica, L. epiphylla, and L. maculata (Grolle, 1976;Pócs & Lye, 1999;Schuster, 1980;Zhu & So, 2001). Leptolejeunea is characterized by its minute size, deeply bilobed underleaves with two widely divergent and subulate lobes, elliptic to narrowly obovate leaf lobes often with dentate margins, the presence of one to several ocelli in leaf lobes, and vegetative reproduction by cladia (Figure 1). Several species show a tendency for dry leaves to become elevated and produce monoterpenes that emit a strong fragrance (Gradstein, Churchill, & Salazar-Allen, 2001) meaning the genus can be readily identified even in the field; yet identification of species is notoriously difficult. Söderström et al. (2016) accepted 48 species but indicated knowledge problems or serious doubts about the taxonomic value of many. An earlier study estimated global diversity at 25 species (Gradstein et al., 2001). So far, only a few accessions have been included in molecular phylogenetic studies (Ahonen, Muonen, & Piippo, 2003;Heinrichs et al., 2014;. Results from these studies rejected a previously hypothesized close relationship between Leptolejeunea and Drepanolejeunea based on shared underleaf shape and the presence of ocelli in leaves of both genera (Gradstein, 2013), and resolved Leptolejeunea in a relatively isolated position within Lejeuneaceae subf. Lejeuneoideae . Lejeuneaceae subtribe Leptolejeuneinae was established as a result to accommodate Leptolejeunea .
However, molecular phylogenetic investigations conducted to date have not improved current morphology-based species concepts nor resolved biogeographic patterns.
Currently, in contradiction to more traditional views of morphological species, widespread Leptolejeunea species are believed to be the result of recent LDD out of Asia, a hypothesis promoted by Schuster (1983: 618): "taxa such as Leptolejeunea elliptica… have shown dispersal, clearly in geologically recent times, well out from Asia into the Pacific, to South America, Central and southern North America." Here, we extend the sampling of Heinrichs et al. (2014) and test previous hypotheses on origins and extant distribution of Leptolejeunea species.
We provide evidence for a Cenozoic origin of the Leptolejeunea crown group and reject pantropical species ranges.
The mix was filled up with double-distilled water to a total volume of 50 μL. Representatives of Pycnolejeunea and Xylolejeunea were chosen as outgroups following phylogenetic hypotheses of , Bechteler, Lee, Schäfer-Verwimp, Pócs, et al. (2016) and .

Malaysia
and an average standard deviation (SD) of split frequency below 0.01 indicated a sufficiently long run. The initial 25% of sampled trees were discarded as burn-in. The remainder were summarized with  (Larget & Simon, 1999).

| Divergence time estimates and biogeography
Dating analyses were performed using BEAST 1.8.2 (Drummond et al., 2012) using the same partitioning scheme and substitution models as the MRBayEs analyses. An ultrametric starting tree without time scale was generated by setting the ingroup monophyletic, using linked trees over all partitions, 60 million generations and sampling every 6,000 generations. An uncorrelated log-normal (UCLN) relaxed clock and a birth-death prior accounting for incomplete sampling (Stadler, 2009) were used. This tree was used as a starting tree for subsequent divergence time estimates. Again, the ingroup was constrained as monophyletic, trees were linked over all partitions, and this analysis ran for 100 million generations sampling every 10,000 generations. As no Leptolejeunea fossils are known, a plastid genome substitution rate of 5 × 10 −4 subst./sites/my (Palmer, 1991;Villarreal & Renner, 2012) (Ree & Smith, 2008), DIVA (dispersalvicariance analysis; Ronquist, 1997), and BayArea (Landis, Matzke, Moore, & Huelsenbeck, 2013), each of which can be extended with an additional free parameter j accounting for founder-event speciation.
To obtain the recommended operational taxonomic units consisting of monophyletic populations and not individual specimens, specimens of one species with the same putative area of endemism were merged together into a single terminal using the R-script provided on the BiogEoBEARS webpage (http://phylo.wikidot.com/example-biogeobears-scripts#pruning_a_tree). All six models were compared using likelihood values, the AIC, and the AIC corrected for small sample size (AIC c ) (Matzke, 2014). The maximum number of areas was set to three to account for the assumed pantropical ranges of Leptolejeunea species (Grolle, 1976;Pócs, 2012;Pócs & Lye, 1999;Schuster, 1983).

| Morphological investigation
Specimens were studied under a Carl Zeiss AxioScope A1 compound microscope equipped with a Canon 60D digital camera using transmit-

| Divergence time estimates and biogeography
The divergence time analyses (Figure 3
Asian L. maculata s.str. is placed in main clade III, together with a Paleotropical lineage here identified as L. schiffneri. Neotropical accessions of L. maculata belong to main clade II and have been identified as L. convexistipa. Such findings have frequently been explained as instances of cryptic or near cryptic speciation (Shaw, 2001); however, molecular topologies may allow revision of morphological evidence and the identification of morphological character states supporting the different lineages (Forrest et al., 2011;Heinrichs et al., 2015;Renner et al., 2013). Revision of Leptolejeunea specimens is challenging as the taxonomy of this genus relies heavily on the number and distribution of ocelli in the leaves, that is, specialized cells containing only a single large rather than several small oil bodies (He & Piippo, 1999). These often disappear from herbarium specimens.
Exceptionally large or small leaf cells in herbarium specimens may be indicative of ocelli; however, ocelli sharing the size of the surrounding leaf cells may not be recognizable in dried materials. A thorough revision of Leptolejeunea thus needs to be based on the investigation of living plants from all parts of the range and sequencing of a comprehensive number of specimens including types or topotypes. New sources of species circumscribing characters also need to be sought.
Such work is beyond the scope of this study; however, our data facilitate discrimination between alternative interpretations of species circumscription and to reconstruct the distribution of the main clades.
Our data also support the finding of Renner (2015) that morphologically similar leafy liverworts may be placed in different main lineages, despite considerable morphological overlap. Accessions originally assigned to the same species were resolved in different main clades, and the supposedly closely related species L. brasiliensis and L. elliptica (Schuster, 1980) were resolved in main clade I or II (Figure 2).
Phylogenies of Lejeuneaceae genera often show a geographical pattern related to the distribution of lineages rather than a morphological pattern. Examples include the genera Lejeunea (Heinrichs et al., 2013) and Diplasiolejeunea (Dong et al., 2012) which exhibit separation into predominantly Neotropical and predominantly Paleotropical lineages. A similar situation manifests in Leptolejeunea.

| Divergence time estimates, biogeography, and infraspecific variation
Our divergence time estimates suggest Cenozoic diversification of Leptolejeunea and contradict Gondwanan vicariance (Raven & Axelrod, 1974) as an explanation for the observed disjunctions. Establishment of the Leptolejeunea crown group in the Eocene accordances well with the appearance of humid megathermal angiosperm forests (Morley, 2011) which provided the preferred epiphyllous habitat of extant Leptolejeunea representatives. Cretaceous gymnosperm forests differed in structure and evaporated less water than tropical angiosperm forests (Boyce & Lee, 2010). Thus, they may not have hosted as diverse epiphyll communities or supported Lejeuneaceae representatives adapted to other niches than modern species . Similar evolutionary processes have been reconstructed for the genera Lejeunea, Harpalejeunea, and Microlejeunea based on molecular and fossil evidence .
Our reconstruction failed to unambiguously identify the area of origin of Leptolejeunea; however, we need to consider the wide distribution of humid angiosperm forests in the Eocene including the northern "boreotropical" region (Morley, 2011 (Lewis, Rozzi, & Goffinet, 2014;Shaw et al., 2010) and leptosporangiate ferns Wei et al., 2015). Liverworts have dispersed to the African continent and associated islands from both the Neotropics and Asia (Feldberg et al., 2007;Heinrichs et al., 2005). Both biogeographical analyses (Figures 3 and   4) provide evidence for an Asian origin of the African accessions of L. epiphylla, L. schiffneri, and L. cf. subrotundifolia, whereas the origin of the African L. astroidea remains unclear. African taxa nesting in Asian clades have also been described for ferns (Hennequin, Hovenkamp, Christenhusz, & Schneider, 2010;Janssen, Kreier, & Schneider, 2007) and angiosperms (Kulju, Sierra, Draisma, Samuel, & van Welzen, 2007;Li, Dressler, Zhang, & Renner, 2009;Richardson, Chatrou, Mols, Erkens, & Pirie, 2004). Monsoon trade winds were proposed as dispersal agent from Asia to Africa (Li et al., 2009) and could also be responsible for the observed pattern in Leptolejeunea. Alternatively, animal-mediated dispersal may contribute to current disjunctions. At small spatial scales, millipedes have been demonstrated to move gemmae of species of the moss genus Calymperes (Zona, 2013). Larger animals that move over correspondingly larger spatial scales may also transport propagules and plant fragments (Lewis et al., 2014). In New Zealand, the isolated occurrences of the tropical Calymperes tenerum are congruent with known visitation sites of the predominantly tropical black-winged petrel (P. J. de Lange, personal communication).
Seabirds are known to visit potential or actual breeding sites, even though visiting individuals may not nest there. To visit these sites, which are often forested, birds literally crash through the canopy to the ground, thus coming into close, vigorous contact with leaf and twig surfaces, providing ample opportunity for plant fragments to become deeply embedded within the bird's feather matrix. Seabirds roam widely during their nonbreeding season and routinely traverse oceans and have been known to traverse land masses bridging oceanic waterways.
The island occurrences provide evidence for the ability of Leptolejeunea species to disperse over long distances either by vegetative propagules (Laenen et al., 2016) or by spores (Van Zanten & Gradstein, 1988). However, successful LDD seems rare in Leptolejeunea, as indicated by the plurispecies clades being restricted to either the Neotropics or the Paleotropics, but also by the genetic variation within single morphospecies. Although our data support a narrower species concept and reinstatement of several putative synonyms, some species clades still have a considerable molecular variation, with initial splits in the late Miocene (Figure 3). Examples include a split between mainland South American L. exocellata and accessions from the West Indian Islands, splits within Asian L. epiphylla, and splits within Neotropical L. convexistipa. Considerable molecular variation related to a geographical rather than a morphological pattern has been observed for a larger number of liverworts (Fuselier et al., 2009;Heinrichs et al., 2015;Ramaiya et al., 2010) although it is still somewhat unclear whether this variation is in general indicative of genetically independent entities. Follow-up studies should thus involve denser sampling and additional markers including microsatellites.
Intercontinental gene flow has already been demonstrated for bryophytes, especially for holarctic species of the moss genus Sphagnum (Kyrkjeeide, Hassel, Flatberg, Shaw, Brochmann, et al., 2016;Shaw et al., 2014); however, the epiphyllous habitat of Leptolejeunea species in the understory of tropical forests may lower the LDD success rate compared to Sphagnum species which occur in open wetland systems.

| Perspectives
Every disjunction has its first day; hence, we cannot generally reject intercontinental or even pantropical species ranges (Lewis et al., 2014). On the other hand, a growing body of evidence indicates that LDD occurs only infrequently in bryophytes and that it is thus often associated with speciation. The accumulation of genetic disparity in bryophytes is often not associated with the accumulation of a similar amount of morphological disparity (Baczkiewicz & Buczkowska, 2016;Ramaiya et al., 2010), although there are exceptions (Heinrichs, Gradstein, Groth, & Lindner, 2003). Lack of molecular support for morphology-based supraspecific taxa such as sections and subgenera (Devos, Renner, Gradstein, Shaw, & Vanderpoorten, 2011) further complicates the understanding of bryophyte evolution and appropriate choice of ingroup representatives. A reliable reconstruction of the evolutionary history and biogeography of bryophytes thus needs to be based on comprehensive molecular phylogenies with complete population-level sampling. Only such phylogenies will facilitate species identification and refined estimation of bryophyte global diversity and origins.